Trophic amplification: A model intercomparison of climate driven changes in marine food webs

increases. By analysing the biomass response within food webs in available MEMs, we found that model parameters and structures contributed to more complex responses than a consistent amplification of climate impacts of higher trophic levels. Our study provides addi- tional insights into the ecological mechanisms that will impact marine ecosystems, thereby informing model and scenario development. 1. Introduction Greenhouse gas emissions from anthropogenic activities have disrupted the natural carbon cycle [1], impacting all ecological compartments of the biosphere [2]. As a result of these changes in physical and chemical properties, major changes have already been observed at the base of the food web and are expected to be amplified in the coming decades, often negatively, at higher trophic levels—a mechanism known as trophic amplification [2–5]. Phytoplankton, which account for 90% of net primary production (NPP) and fuel marine food webs [6, 7], are affected by climate-driven changes in biomass and productivity through changes in nutrient availability, light limitation and thermal stratification of the water column [8]. Earth System Models (ESMs) within the 6th Coupled Model Intercomparison Project (CMIP6; [9]) project an overall global decrease of 3.0 ± 9.1% in NPP by the end of the 21st cen- tury [10–13]. However, the ESMs also project large spatial variability in NPP changes. Ocean warming, deoxygenation and acidification affect the physiological functions of marine organisms, including modifications in body function, growth rates, maximum body

size and reproductive rates [4, 14–17]. Moreover, climate-induced changes in ocean conditions affect the biogeography and phenology of marine populations [2], consequently altering the trophodynamics of marine ecosystems [3, 18, 19]. These multiple climatic stressors may inter- act and amplify the impacts on marine ecosystems [4, 20, 21]. Changes in biomass and productivity at low trophic levels may be exacerbated at higher tro- phic levels through trophic amplification. This process refers to the bottom-up propagation of the climate signal from primary producers along the food web, with an increasing magnitude of impacts at each higher trophic level [3, 22–25]. Trophic amplification between phytoplank- ton and zooplankton has been demonstrated in simulations from different planktonic food web models and ESMs [22, 24, 26], with changes in zooplankton biomass around 2 times greater than those of phytoplankton. Trophic amplification has also been explored at higher trophic levels, with regional-scale studies showing that indirect effects of increased temperature can be exacerbated through the food web, leading to greater impacts on the biomass of higher trophic levels [15]. Lotze et al. (2019) analysed trophic amplification based on a model intercomparison project using six global FishMIP Marine Ecosystem Models (MEMs) forced by standardised outputs of two contrasting ESMs and four climate change scenarios. They defined trophic amplification by comparing relative changes in total consumer biomass (MEMs outputs) to relative changes in

NPP and low trophic level biomass (ESM outputs). They showed that, by the end of the 21st century, the total consumer biomass was projected to decline more strongly than NPP, but with high levels of uncertainty. The nature of temperature-dependence of biological rates plays a role in the transfer effi- ciency of biomass or energy from one trophic level to the next (or consumer in a more com- plex food-web) and MEMs represent the details of these processes differently [25, 27]. Through trophic amplification, high trophic level species are hypothesised to be even more PLOS ONE Trophic amplification revealed PLOS ONE | https://doi.org/10.1371/journal.pone.0287570 August 23, 2023 2 / 23 funding support from the Region Bretagne. Hubert du Pontavice, William W. L. Cheung, and Gabriel Reygondeau acknowledge funding support from the Nippon Foundation-UBC Nereus Program. JDE acknowledges funding support from Australian Research Council Discovery Project DP190102293. CMP acknowledges funding support from NOAA Grants NA20OAR4310438, NA20OAR4310441 and NA20OAR4310442. JG acknowledges funding support from NASA Grant 80NSSC21K0420. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.